Soil errors in legacy data: Mg truncated
Email from Bob 1/25/2007:
Elizabeth Marx discovered that soils for plot 20-7-340-8 have mg ppm as 2,
which is inconsistent with the nature of the soils, to say nothing of the
high percent bases associated with mg. I checked old files and this seems
to have been there all along. Puzzled, I found the old printouts and
discovered that throughout, the data contain only the rightmost 3 digits
for Mg. A fourth digit is rare but does sometimes occur. In Elizabeth's
case the value had been 2002 and just 002.
It should be possible to back calculate based on base saturation % to
identify which records contain this left truncation error. I have no idea
how many projects this is a problem for, so once we have the test, we
should run it on all soils data in CVS.
Would you all please work this out for me.
#1 Updated by Michael Lee almost 15 years ago
Does this problem possibly apply to other fields besides Mg? What kind of range
of Basal Saturation are we looking for, coupled with how low an Mg?
Do we need to consult with the original printouts anyway, as how would
be know if "2" should be 2002 or 1002 or 3002?
#2 Updated by Michael Lee almost 15 years ago
From Bob's email 1/26/2007:
I have not yet gone back to check, but it could be that the way the
the total exchangeable bases is derive is through calculation based on Ca,
Mg, K etc, in which case given we know TEC we could calculate Mg. I leave
this to Forbes to work out for now.
#4 Updated by Michael Lee over 12 years ago
This bug persists in the new archive database. Plot 20-7-430 still has Mg of 2 for one reading:
authorObsCode EC Mg_ppm percent_Mg
020-07-0340 40.17 493 51.72
020-07-0340 25.52 280 74.45
020-07-0340 23.79 2 70.13
020-07-0340 34.96 436 81.9
020-07-0340 24.97 98 70.02